High-dimensional electrical stimulation for visual prosthesis

ongoing

Future advanced neuroprostheses will need to transfer orders of magnitude more information to the brain than currently possible. This is most urgently needed in visual prostheses. Improving the electrode count will be part of the solution: a next generation of visual prosthesis will most probably be based on the insertion of over 1000 microelectrodes in the visual cortex. Still, current visual prostheses use very simple stimulation patterns, in which at most the stimulation amplitude is modulated. We propose to explore a second, complementary approach to brute scaling: using the available electrodes more efficiently by applying sophisticated stimulation protocols. Our main objective is to achieve a fundamental breakthrough in the spatial resolution of electrical brain stimulation to restore vision, obtaining a resolution of at least 20X the number of electrodes that are physically present. The vast number of possible stimulation combinations calls for a radically new research methodology, integrating modeling and state-of-the-art neuroscience methods at every spatial scale (from single neurons to the entire brain) in a closed-loop optimization process. With this combination of techniques, we will study which stimulation patterns effectively induce sufficient neural activations in higher areas (i.e. ignition) and cause visual perceptions. Thus, we will be able to explore the vast, hyperdimensional search space of possible stimulation patterns, and produce a set of in vivo tested stimulation patterns that are capable of eliciting distinguishable physiological and behavioral responses. The obtained order-of-magnitude improvement in resolution will spur the development of breakthrough prostheses that will be widely adopted by blind patients, and bring the field of neural interfacing to the next level.